R Markdown

Simpson’s Paradox and Revealing the truth about Lurking Variables

In 1973, UC Berkeley found a startling discrepancy while looking into their acceptance rates between male and female applicants. The claim dived into females getting rejected more than males. Females were applying to more difficult departments resulting in more rejection.

The story brings along a problem towards their method which is the data set most likely to have more than one graph to display their procedure. Looking more into each department, the female and male acceptance rate proves otherwise.

Shown in the graph above, each department except for department A are relatively close in status. The problem at the beginning of the story was females getting rejected at a higher rate but in department A. This exposes the lurking variable because department A reveals the actual opposite of the claim. In this representation, males was the gender who got rejected more than females.

This is a great example of Simpsons paradox which shows a specific trend within the data but once another variable is introduced the trend disappears or even gets reversed.